Those seriously interested in the meaning and the politics of racial epithets (as some of the commenters on Pakigate, Sootygate, Gollygate seem to be) should take a look at a paper called "The semantics of racial epithets", published by Christopher Hom in The Journal of Philosophy CV [= 105], no. 8 (August 2008), pp. 416-440. This is a technical paper in philosophical semantics (it's philosophy, not linguistics; and let me say that I do not necessarily endorse the view that it defends). Hom outlines its aim on his website thus:
Racial epithets are derogatory expressions, understood to convey contempt toward their targets. But what do they actually mean, if anything? While the prevailing view is that epithets are to be explained pragmatically, I argue that a careful consideration of the data strongly supports a particular semantic theory. I call this view Combinatorial Externalism (CE). CE holds that epithets express complex properties that are determined by the discriminatory practices and stereotypes of their corresponding racist institutions. Depending on the character of the institution, the complex semantic value can be composed of a variety of components. The account has significant implications on theoretical, as well as, practical dimensions, providing new arguments against radical contextualism, and for the exclusion of certain epithets from First Amendment speech protection.
Thus Hom is offering a reasoned case that it is best to see the denigratory character of racial epithets as built into their actual conventional meanings, and not just as a possible concomitant of some of their occasional uses. (Many of commenters seem to align with this view, though they tend to just assert it and call any other views absurd, rather than present arguments.)
At the beginning of the paper, Hom sets out his aims thus:
There are two competing strategies for explaining how epithets work, one semantic and the other pragmatic. According to the semantic strategy, their derogatory content is fundamentally part of their literal meaning, and thus gets expressed in every context of utterance. This strategy honors the intuition that epithets literally say bad things, regardless of how they are used. According to the pragmatic strategy, their derogatory content is fundamentally part of how they are used, and results from features of the individual contexts surrounding their utterance. This strategy honors the intuition that epithets can be used for a variety of purposes, and that this complexity surrounding epithets precludes a univocal, context-independent explanation for how they work. Neither view is without difficulty, although to many the pragmatic strategy is prima facie more attractive. I shall argue, however, that the semantic strategy actually fares better on a number of criteria. In doing so, I shall motivate a particular semantic account of epithets that I call combinatorial externalism. The account has significant implications on theoretical, as well as, practical dimensions.
In something of a departure for Language Log, which generally deals with material less technical than Hom's paper, I open the comments space below for use by people who have read the paper, please, not just for opinions stimulated by the above summary, or (pretty please!) random observations about racism or nasty words. I'll enforce the policy to the extent that time permits.